A lateral reference position decision method for autonomous driving path planning and vehicle
By introducing a lateral reference position decision method, the problems of high computational load and path instability in autonomous driving path planning are solved, achieving efficient and safe path planning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NEOLIX TECH CO LTD
- Filing Date
- 2024-04-30
- Publication Date
- 2026-07-21
AI Technical Summary
In existing autonomous driving path planning, the decoupling of lateral and longitudinal control leads to high computational load, low efficiency, unstable path generation, and difficulty in coping with complex changes in the external environment.
A lateral reference position decision method is introduced. By acquiring lane boundary and obstacle information, a Frenet coordinate system is constructed, lateral limit positions are analyzed, an initial lateral reference position is generated, and filtering is performed to ensure that the path is within the lane and avoids obstacles.
Reduce the computational load of path planning, improve vehicle driving efficiency and safety, and ensure path stability and safety.
Smart Images
Figure CN118329063B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of vehicle autonomous driving planning technology, specifically to a lateral reference position decision method and vehicle for autonomous driving path planning. Background Technology
[0002] Currently, autonomous driving planning for vehicles typically employs a decoupled lateral and longitudinal control model. Lateral control primarily manages the vehicle's position within the lane, while longitudinal control controls its speed. Lateral control involves path planning in the Frenet coordinate system. The initial step in path planning requires finding a feasible path within the topological space formed by obstacles and roads, for example, by constructing a graph structure and using algorithms like A* to find an optimal topological space or path. In traditional autonomous driving control, the vehicle's route is usually generated after complex calculations based on various dynamic parameters, including road reference lines, vehicle motion state, obstacle information, and vehicle parameters. This results in excessive computation, low efficiency, poor real-time route generation, and, due to the variability and complexity of the external environment, unstable paths, which is detrimental to vehicle control. Summary of the Invention
[0003] To overcome the shortcomings of the prior art, this application provides a lateral reference position decision method and vehicle for autonomous driving path planning, specifically adopting the following technical solution:
[0004] A lateral reference position decision method for autonomous driving path planning includes the following steps:
[0005] The driving boundaries and offset decisions within the lane are obtained, wherein the driving boundaries include at least the left boundary, the right boundary, and the centerline; and the offset decisions include at least offset driving decisions and non-offset driving decisions.
[0006] A Frenet coordinate system is constructed based on the centerline of the lane, with the position of the centerline in the lateral direction as 0, the left side of the centerline as the positive direction, and the right side of the centerline as the negative direction;
[0007] The left and right boundaries of the driving boundary are discretized longitudinally to obtain several longitudinally arranged boundary point pairs, each of which includes a left boundary point and a right boundary point.
[0008] Traverse each pair of boundary points vertically to obtain the lateral limit positions corresponding to different pairs of boundary points. The lateral limit positions include the left limit position and the right limit position.
[0009] The initial lateral reference position is obtained based on offset decision and lateral limit position analysis;
[0010] Define the obstacle detection boundaries and obstacle containers for the vehicle, traverse obstacles in the longitudinal direction and classify them;
[0011] Based on the obstacle classification results, obstacle avoidance decisions are made for obstacles at different locations in the longitudinal direction, and lateral reference positions are calculated.
[0012] The lateral reference positions at different locations are filtered to obtain the filtered lateral reference positions.
[0013] The method of this application introduces a lateral reference position for vehicle path planning, and the final driving path generated by the vehicle conforms to this lateral reference position as closely as possible, which can ensure safe and stable driving of the vehicle. This method reduces the influence of some dynamic parameters in the path planning process, reduces the computational load of path planning, and improves vehicle driving efficiency.
[0014] Optionally: The step of obtaining the initial lateral reference position based on offset decision and lateral limit position analysis includes:
[0015] When making an off-center driving decision, obtain the centerline of the lane and iterate through the curvature at each position on the centerline;
[0016] The type of lane ahead of the vehicle is determined based on the curvature of the centerline, and the lane type includes at least a curve.
[0017] When it is determined that there is a curve in front of the vehicle, the first reference position is calculated based on the right limit position and the preset limit margin; the position of the first reference position and the center line are determined, and the position that is more inclined to the right boundary point is selected as the initial lateral reference position.
[0018] Optionally: The step of obtaining the initial lateral reference position based on offset decision and lateral limit position analysis includes:
[0019] When making an off-center driving decision, obtain the centerline of the lane and iterate through the curvature at each position on the centerline;
[0020] The type of lane in front of the vehicle is determined based on the curvature of the centerline, and the lane type includes at least a straight road;
[0021] When it is determined that the road in front of the vehicle is straight, check whether there are obstacles at the right boundary points of the lane;
[0022] When there is an obstacle at any right boundary point, the first reference position is calculated based on the right limit position and the preset limit margin corresponding to the right boundary point.
[0023] When the first reference position is located between the right limit position and the left boundary point in the same vertical position, the initial horizontal reference position is selected from the first reference position;
[0024] When the first reference position is not located between the right limit position and the left boundary point in the same vertical position, the initial horizontal reference position is selected as the position that is closest to the first reference position between the right limit position and the left boundary point.
[0025] When there are no obstacles at the right boundary points of the lane, the first reference position is calculated based on the right limit position and the preset limit margin.
[0026] Determine the position of the first reference position relative to the center line. The initial lateral reference position is selected as the position closest to the right boundary point between the first reference position and the center line.
[0027] Optionally: The step of obtaining the initial lateral reference position based on offset decision and lateral limit position analysis includes:
[0028] When a non-offset driving decision is executed, and there are obstacles at both the right and left boundary points of the same longitudinal position of the lane, the lateral distance between the obstacles corresponding to the right and left boundary points and the vehicle is determined respectively.
[0029] Obtain the obstacle corresponding to the boundary point closest to the vehicle, and obtain the lateral limit position on the side where the obstacle is located;
[0030] The first reference position is calculated based on the lateral limit position on the side where the obstacle is located and the preset limit margin.
[0031] Analyze the first reference position, the lateral limit position corresponding to the boundary point on the opposite side of the obstacle, and the lateral coordinates of the centerline;
[0032] When the first reference position is located between the lateral limit position on the opposite side of the obstacle and the centerline, the initial lateral reference position is selected from the first reference position;
[0033] When the first reference position is not located between the lateral limit position and the center line on the opposite side of the obstacle, the initial lateral reference position is selected as the position closest to the first reference position between the lateral limit position and the center line on the opposite side of the obstacle.
[0034] Optionally: The step of obtaining the initial lateral reference position based on offset decision and lateral limit position analysis includes:
[0035] When a non-offset driving decision is executed, and there is an obstacle at any boundary point of the right or left boundary point at the same longitudinal position of the lane, the first reference position is calculated based on the lateral limit position corresponding to the boundary point on the side where the obstacle is located and the preset limit margin.
[0036] Analyze and determine the first reference position, the lateral limit position corresponding to the boundary point on the side where the obstacle is located, and the lateral coordinates of the centerline;
[0037] When the first reference position is located between the lateral limit position on the side where the obstacle is located and the center line, the initial lateral reference position is selected as the first reference position;
[0038] When the first reference position is not located between the lateral limit position and the center line on the side where the obstacle is located, the initial lateral reference position is selected as the position closest to the first reference position between the lateral limit position and the center line on the side where the obstacle is located.
[0039] Optionally: The step of obtaining the initial lateral reference position based on offset decision and lateral limit position analysis includes:
[0040] When a non-offset driving decision is executed, and there are no obstacles at the left and right boundary points of the same longitudinal position of the lane, the distances between the right and left boundary points and the vehicle are determined respectively.
[0041] If the distance between the left and right boundary points at the same longitudinal position of the lane is less than a preset threshold, the initial lateral reference position is selected as the middle position between the current left and right boundary points;
[0042] If the sum of the lateral coordinates of the right boundary point of the lane and the preset road margin is greater than 0, the initial lateral reference position is selected as the lateral position corresponding to the sum of the lateral coordinates of the right boundary point of the lane and the preset road margin.
[0043] If the difference between the lateral coordinate of the left boundary point of the lane and the preset road margin is less than 0, the initial lateral reference position is selected as the lateral position corresponding to the difference between the lateral coordinate of the left boundary point of the lane and the preset road margin.
[0044] Otherwise, the initial lateral reference position is selected from the lateral position where the center line is located.
[0045] The above process analyzes static obstacles based on factors such as offset decision and lateral limit position to obtain an initial lateral reference position. This initial lateral reference position can be used as a reference factor for subsequent dynamic obstacle analysis, thereby improving the efficiency of driving path planning.
[0046] Optionally: The step of setting obstacle detection boundaries for the vehicle includes:
[0047] Obtain the vehicle's own body boundary and the road boundary of the lane; the vehicle body boundary includes the front boundary, rear boundary, left boundary and right boundary, and the road boundary includes the left road boundary and right road boundary;
[0048] The front obstacle detection boundary of the vehicle is calculated based on the front boundary of the vehicle and the preset forward detection distance.
[0049] The rear obstacle detection boundary of the vehicle is calculated based on the rear boundary of the vehicle and the preset rearward detection distance;
[0050] The first detection position is calculated based on the left boundary of the vehicle and the preset left detection distance, and the second detection position is calculated based on the left road boundary and the preset left road boundary margin; the position that is biased towards the center line between the first detection position and the second detection position is selected as the left obstacle detection boundary of the vehicle.
[0051] The third detection position is obtained by calculating the right boundary of the vehicle and the preset right detection distance, and the fourth detection position is obtained by calculating the right road boundary and the preset right road boundary margin. The position that is biased towards the center line between the third detection position and the fourth detection position is selected as the right obstacle detection boundary of the vehicle.
[0052] The obstacle detection boundary of the vehicle is obtained by merging the obtained front obstacle detection boundary, rear obstacle detection boundary, left obstacle detection boundary and right obstacle detection boundary.
[0053] Optionally: The step of setting the obstacle detection boundaries and obstacle containers for the vehicle, traversing longitudinal obstacles, and classifying them includes:
[0054] Create empty left and right obstacle containers respectively;
[0055] Traverse vertically the obstacles located within the obstacle detection boundary of the vehicle and obtain the boundaries of the obstacles;
[0056] When the lateral coordinate of the left boundary of the obstacle is greater than the lateral coordinate of the vehicle's center axis, and the lateral coordinate of the right boundary of the obstacle is less than the lateral coordinate of the vehicle's center axis, the obstacle is simultaneously placed into the left obstacle container and the right obstacle container.
[0057] When the lateral coordinate of the left boundary of the obstacle is between the lateral coordinate of the right obstacle detection boundary and the lateral coordinate of the vehicle's center axis, the obstacle is placed into the right obstacle container.
[0058] When the lateral coordinate of the right boundary of the obstacle is located between the lateral coordinate of the left obstacle detection boundary and the lateral coordinate of the vehicle's center axis, the obstacle is placed into the left obstacle container.
[0059] Optionally: The step of matching and avoiding obstacles at different positions in the longitudinal direction based on the obstacle classification results and calculating the lateral reference position includes:
[0060] When both the left and right obstacle containers of the vehicle are filled with obstacles, the lateral reference position is selected from the lateral coordinates of the vehicle's center axis.
[0061] When neither the left nor right obstacle container of the vehicle is occupied by an obstacle, the lateral reference position is selected as the initial lateral reference position of the current lane.
[0062] When only the left obstacle container contains an obstacle, traverse the left obstacle container and obtain the right boundary of the obstacle closest to the vehicle. Based on the right boundary of the obstacle, extend to the right by a first preset margin to obtain the first avoidance boundary; based on the right road boundary of the obstacle's current position, extend to the left by a second preset margin to obtain the second avoidance boundary.
[0063] The position that is biased towards the obstacle between the first avoidance boundary and the second avoidance boundary is selected as the first decision avoidance boundary.
[0064] The position furthest from the obstacle among the vehicle's centerline, the initial lateral reference position, and the first decision avoidance boundary is selected as the lateral reference position.
[0065] When only the right obstacle container contains an obstacle, traverse the right obstacle container and obtain the left boundary of the obstacle closest to the vehicle. Based on the left boundary of the obstacle, extend to the left by a third preset margin to obtain the third avoidance boundary; based on the left road boundary of the obstacle's current position, extend to the right by a fourth preset margin to obtain the fourth avoidance boundary.
[0066] The position that is biased towards the obstacle between the third and fourth avoidance boundaries is selected as the second decision avoidance boundary;
[0067] The position furthest from the obstacle among the vehicle's centerline, the initial lateral reference position, and the second decision avoidance boundary is selected as the lateral reference position.
[0068] This application classifies and analyzes obstacles by using obstacle detection boundaries and obstacle containers, and combines the obtained initial lateral reference position to provide a reference for vehicle dynamic driving, and generates the final lateral reference position. By controlling the vehicle's driving trajectory to conform to the lateral reference position, it can ensure that the vehicle safely avoids obstacles, improve vehicle driving safety, and improve path planning efficiency.
[0069] In addition, this application also discloses a vehicle comprising: a controller configured to perform the lateral reference position decision method described above.
[0070] Beneficial effects
[0071] The technical solution of this application achieves the following beneficial effects:
[0072] The lateral reference position decision method of this application introduces a lateral reference position as a reference quantity for vehicle path planning. This lateral reference position includes at least the following characteristics: ensuring the driving path remains within the lane, ensuring the driving path travels at a defined lateral position, ensuring the driving path maintains continuous frames with minimal abrupt changes, and ensuring the driving path can avoid dynamic and static obstacles. When the lateral reference position generated based on these characteristics is applied to vehicle path planning, it is only necessary to ensure that the final generated driving path conforms as closely as possible to this lateral reference position to guarantee safe and stable vehicle driving. When using the method of this application for path planning during vehicle operation, there is no need to consider complex calculations of various related parameters such as road reference lines, vehicle motion state, obstacle information, and vehicle's own parameters, which reduces the computational load of path planning and improves planning efficiency. Attached Figure Description
[0073] Figure 1 This is a flowchart illustrating the lateral reference position decision method in an embodiment of this application.
[0074] Figure 2 This is a schematic diagram illustrating the planning of driving boundaries under different conditions in the embodiments of this application. Figure 2 (1) is the boundary state diagram for the cruise phase; Figure 2 (2) is a boundary state diagram for detouring when there is an obstacle in the lane; Figure 2 (3) is the extended boundary state diagram when borrowing a passage; Figure 2 (4) is a boundary state diagram of the adjacent lane during the lane changing process.
[0075] Figure 3 This is a schematic diagram illustrating the process of discretizing the driving boundary and setting the lateral limit position in an embodiment of this application. Figure 3 (1) is the state diagram of the discretized driving boundary; Figure 3 (2) is the state diagram of the boundary point pair of the driving boundary; Figure 3 (3) is a state diagram for setting the horizontal limit position. Detailed Implementation
[0076] The present application will now be further described with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present application and should not be construed as limiting the scope of protection of the present application. It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present application.
[0077] Currently, autonomous driving planning for vehicles typically employs a decoupled lateral and longitudinal control model. Lateral control primarily manages the vehicle's position within the lane, while longitudinal control controls its speed. A suitable route is generated by combining lateral and longitudinal control. In traditional autonomous driving control, the vehicle's route is generated through complex calculations based on various parameters, including road reference lines, vehicle motion state, obstacle information, and vehicle parameters. This results in excessive computation, low efficiency, poor real-time route generation, and instability due to the variability and complexity of the external environment, hindering vehicle control. This application introduces the concept of a "lateral reference position," which is essentially a lateral coordinate in the Frenet coordinate system, represented by a floating-point number. This value remains constant within each planning frame and is updated with each path planning iteration. The lateral reference position introduced in this application ensures the driving path remains within the lane, operates at a defined lateral position, maintains a continuous and relatively stable driving path across frames, and avoids both static and dynamic obstacles. The lateral reference position generated based on the above principles, when applied to vehicle lateral decision-making, only needs to ensure that the final generated driving path conforms as closely as possible to this lateral reference position to guarantee safe and stable vehicle driving. Therefore, after obtaining a desired lateral reference position, this application can apply it to the cost design of graph result nodes and edges in the autonomous driving control process, thereby generating a driving path that conforms to this lateral reference position.
[0078] like Figure 1 As shown, this application specifically discloses a lateral reference position decision method for autonomous driving path planning, which includes the following steps:
[0079] Step 1:
[0080] The driving boundaries and offset decisions within the lane are obtained, wherein the driving boundaries include at least the left boundary, the right boundary, and the centerline; and the offset decisions include at least offset driving decisions and non-offset driving decisions.
[0081] In this application, the driving boundary is the lateral decision topology space, which indicates the driving limits that vehicles are allowed to travel within the lane, such as... Figure 2As shown in the diagram, the thick solid lines representing the lane boundary and obstacle boundary during the cruising phase, the extended boundary when borrowing lanes, and the side lane boundary during lane changing are the left and right boundaries. Furthermore, a center line is provided in the middle of the driving boundaries; this center line serves as a basic reference line during the construction of the Frenet coordinate system. In certain traffic conditions, vehicles may need to deviate from their lane, including but not limited to: left curves ahead, right curves ahead, and intersections. For example, there may be a need to drive on the right side of a non-motorized vehicle lane, or to drive on the right side of a curve; in these situations, the vehicle needs to perform a lane deviation maneuver.
[0082] It should be noted that the method of obtaining the lateral decision topology space in this embodiment is not limited. For example, Apollo's Em planner uses the DP algorithm to open up a convex space as the lateral decision topology space, or it can construct a graph structure based on obstacles, divide the planning space into subspaces as nodes of the graph, define node costs and edge costs according to geometric relationships, and then use algorithms such as Dijkstra's algorithm or A* algorithm to search for a subspace sequence as the lateral decision topology space.
[0083] Step Two:
[0084] A Frenet coordinate system is constructed based on the centerline of the lane, with the position of the centerline in the lateral direction as 0, the left side of the centerline as the positive direction, and the right side of the centerline as the negative direction.
[0085] In this embodiment, the Frenet coordinate system generally uses the centerline of the lane as the reference line. The coordinate system is then established using the tangent vector and normal vector of the reference line. The vertical axis represents the distance the vehicle travels along the centerline, and the horizontal axis represents the distance the vehicle deviates from the centerline, for example, F(s,d), where s represents the vertical axis and d represents the horizontal axis. Since this embodiment takes the left side of the centerline as the positive direction and the right side of the centerline as the negative direction, the value of d in this embodiment distinguishes the orientation of the vehicle relative to the centerline by positive and negative values. For example, F(1,1) indicates that the vehicle is 1m in the longitudinal direction and 1m to the left of the centerline in the lateral direction; F(1,-1) indicates that the vehicle is 1m in the longitudinal direction and 1m to the right of the centerline in the lateral direction.
[0086] Step 3:
[0087] The left and right boundaries of the driving boundary are discretized longitudinally to obtain several longitudinally arranged boundary point pairs, each of which includes a left boundary point and a right boundary point.
[0088] like Figure 3As shown in Figure (1), in this embodiment, the driving boundary is discretized and then divided into several discrete points arranged at intervals along the longitudinal direction. These discrete points can be used as reference points for subsequent positions, reducing the amount of data calculation.
[0089] Step Four:
[0090] Traverse each boundary point pair vertically to obtain the lateral limit positions corresponding to different boundary point pairs. The lateral limit positions include the left limit position and the right limit position.
[0091] For example, in this embodiment, detouring within the lane is taken as an example, such as Figure 3 As shown in (2), obstacles 1 and 2 are placed from near to far from the vehicle. Obstacle 1 is located on the right boundary, and obstacle 2 is located on the left boundary. First, the driving boundary is discretized, and pairs of left and right boundary points (black solid dots) are obtained in the longitudinal direction. Then, the driving boundary is traversed in the longitudinal direction, for example... Figure 3 As shown in (3), the traversal is performed along the right boundary. When the traversal reaches obstacle 1, the right boundary is located to the left of the obstacle. At this time, the right boundary point a, which is closest to the vehicle on the right boundary of the obstacle, is selected. Correspondingly, there is a left boundary point A on the left boundary. Then, in this longitudinal position, the horizontal coordinate of the right boundary point a is taken as the right limit position of the current longitudinal position, and the horizontal coordinate of the left boundary point A is taken as the left limit position of the current longitudinal position. Figure 3 As shown in (3), the traversal is performed along the left boundary. When the traversal reaches obstacle 2, the left boundary is located to the right of the obstacle. At this time, the left boundary point B, which is closest to the vehicle on the left boundary of the obstacle, is selected. Correspondingly, there is a right boundary point b on the right boundary. Then, the lateral coordinate of the right boundary point b is taken as the right limit position of the current longitudinal position, and the lateral coordinate of the left boundary point B is taken as the left limit position of the current longitudinal position. The lateral limit position obtained in this step can provide a basis for the subsequent design of the initial lateral reference position.
[0092] Step 5:
[0093] The initial lateral reference position is obtained based on offset decision and lateral limit position analysis.
[0094] It should be noted that in this embodiment, different analyses are required for different scenarios when the vehicle makes a deviation driving decision and a non-deviation driving decision:
[0095] (1) For the case of executing a deviation driving decision:
[0096] When making a driving offset decision, the centerline of the lane must first be obtained, and the curvature at each position on the centerline must be traversed.
[0097] The type of lane in front of the vehicle is determined based on the curvature of the centerline. In this embodiment, the lane types include at least curves and straight sections.
[0098] For curves:
[0099] When it is determined that there is a curve ahead of the vehicle, a first reference position will be calculated based on the right limit position and a preset limit margin; for example, the first reference position is R. 极限 +buffer, where R 极限 The horizontal coordinate of the right limit position corresponding to the vertical position is given by , and buffer is the custom margin of the right limit position corresponding to the vertical position, which is obtained through historical or empirical data.
[0100] Determine the positions of the first reference point and the centerline, and select the position that is more biased towards the right boundary as the initial lateral reference position ref_l, which is expressed as ref_l = min(0,R). 极限 +buffer), where 0 represents the horizontal coordinate of the centerline of the current vertical position.
[0101] For straight sections:
[0102] When it is determined that there is a straight road ahead of the vehicle, the vehicle makes a deviation driving decision to drive to the right. At this time, the right boundary point of the lane is traversed to check for obstacles.
[0103] When there is an obstacle at any right boundary point, the first reference position is calculated based on the right limit position and the preset limit margin corresponding to the right boundary point.
[0104] When the first reference position is located between the right limit position and the left boundary point in the same vertical position, the initial horizontal reference position is selected from the first reference position;
[0105] When the first reference position is not located between the right limit position and the left boundary point in the same vertical position, the initial lateral reference position is selected as the position that is closest to the first reference position between the right limit position and the left boundary point.
[0106] For example, in this embodiment, when there is an obstacle at the right boundary point, the initial lateral reference position can be represented as ref_l = clamp(R). 极限 +buffer,R 极限 ,L 边界 ), where R 极限 L represents the horizontal coordinate of the right limit position corresponding to the vertical position, and buffer represents the custom margin of the right limit position corresponding to the vertical position. 边界 Here are the lateral coordinates of the left boundary point corresponding to the longitudinal position. The above formula indicates that the initial lateral reference position is preferably located at R. 极限 When selecting the initial horizontal reference position for the +buffer, R must be determined first.极限 Is the horizontal coordinate corresponding to +buffer located at the right limit R of the corresponding vertical position? 极限 With left boundary point L 边界 Between, when it is between the two, choose R. 极限 +buffer; and R 极限 +buffer's horizontal coordinate is less than R 极限 When choosing the horizontal coordinate, select R. 极限 The current position is the initial lateral reference position; when R 极限 +buffer's horizontal coordinate is greater than L 边界 When choosing the horizontal coordinate, select L. 边界 The location is the initial horizontal reference position.
[0107] Furthermore, when there are no obstacles at the right boundary points of the lane, the first reference position is calculated based on the right limit position and the preset limit margin.
[0108] Determine the position of the first reference position relative to the center line. The initial lateral reference position is selected as the position closest to the right boundary point between the first reference position and the center line.
[0109] For example, in this embodiment, for the case where there are no obstacles at the right boundary point, the initial lateral reference position can be expressed as ref_l = min(R). 极限 +buffer,0), where R 极限 Here, R represents the horizontal coordinate of the right limit position corresponding to the vertical position, buffer is the custom margin of the right limit position corresponding to the vertical position, and 0 represents the horizontal coordinate of the center line of the corresponding vertical position. The above formula indicates that when R... 极限 When the horizontal coordinate of +buffer is to the right of the center line, select R. 极限 The current position is the initial lateral reference position, and when R 极限 When the horizontal coordinate of +buffer is to the left of the center line, the position of the center line is selected as the initial horizontal reference position.
[0110] (2) For situations where non-deviation driving decisions are made:
[0111] In detail, in this embodiment, when the vehicle makes a non-deviation driving decision, it analyzes the obstacles located on the left and right driving boundaries respectively:
[0112] A. When a non-offset driving decision is executed, and there are obstacles at both the right and left boundary points of the same longitudinal position of the lane, determine the lateral distance between the obstacles corresponding to the right and left boundary points and the vehicle.
[0113] Obtain the obstacle corresponding to the boundary point closest to the vehicle, and obtain the lateral limit position on the side where the obstacle is located;
[0114] The first reference position is calculated based on the lateral limit position on the side where the obstacle is located and the preset limit margin.
[0115] Analyze the first reference position, the lateral limit position corresponding to the boundary point on the opposite side of the obstacle, and the lateral coordinates of the centerline;
[0116] When the first reference position is located between the lateral limit position on the opposite side of the obstacle and the centerline, the initial lateral reference position is selected from the first reference position;
[0117] When the first reference position is not located between the lateral limit position and the center line on the opposite side of the obstacle, the initial lateral reference position is selected as the position closest to the first reference position between the lateral limit position and the center line on the opposite side of the obstacle.
[0118] For example, in this embodiment, there are obstacles at both the left and right boundary points, and the obstacle at the right boundary point is closer to the vehicle in the longitudinal direction. In this case, the obstacle corresponding to the right boundary point is obtained, and the corresponding right limit position is obtained based on this right boundary point. The initial lateral reference position can then be represented as: ref_l = clamp(R 极限 +buffer,0,L 边界 ), which indicates R 极限 Given that the buffer is located at the centerline and the left boundary point, R is preferred. 极限 The horizontal coordinate of +buffer is the initial horizontal reference position; while when R 极限 When the +buffer position is to the right of the centerline, the position of the centerline is selected as the initial lateral reference position; when R 极限 +buffer position exceeds left boundary point L 边界 When selecting L 边界 The location is used as the initial lateral reference position.
[0119] Similarly, if the obstacle located at the left boundary point is closer to the vehicle in the longitudinal direction, then the obstacle corresponding to the left boundary point is obtained, and the corresponding left limit position is obtained based on the left boundary point. The initial lateral reference position can then be expressed as: ref_l = clamp(L 极限 +buffer,R 边界 ,0), which indicates L 极限 Given that the buffer is located at the horizontal coordinates of the centerline and the right boundary point, L is preferred. 极限 The horizontal coordinate of +buffer is the initial horizontal reference position; while when L 极限 When the +buffer position is to the left of the centerline, the centerline position is selected as the initial lateral reference position; when L 极限+buffer position exceeds right boundary point R 边界 When selecting R 边界 The location is used as the initial lateral reference position.
[0120] B. When a non-offset driving decision is executed, and there is an obstacle at any boundary point of the right or left boundary point at the same longitudinal position of the lane, the first reference position is calculated based on the lateral limit position corresponding to the boundary point on the side where the obstacle is located and the preset limit margin.
[0121] Analyze and determine the first reference position, the lateral limit position corresponding to the boundary point on the side where the obstacle is located, and the lateral coordinates of the centerline;
[0122] When the first reference position is located between the lateral limit position on the side where the obstacle is located and the center line, the initial lateral reference position is selected as the first reference position;
[0123] When the first reference position is not located between the lateral limit position and the center line on the side where the obstacle is located, the initial lateral reference position is selected as the position closest to the first reference position between the lateral limit position and the center line on the side where the obstacle is located.
[0124] For example, in this embodiment, if there is an obstacle at the right boundary point, the corresponding right limit position is obtained based on the right boundary point corresponding to the obstacle. At this time, the initial lateral reference position can be expressed as: ref_l = clamp(R 极限 +buffer,0,L 边界 ), which indicates R 极限 Given that the buffer is located at the centerline and the left boundary point, R is preferred. 极限 The horizontal coordinate of +buffer is the initial horizontal reference position; while when R 极限 When the +buffer position is to the right of the centerline, the position of the centerline is selected as the initial lateral reference position; when R 极限 +buffer position exceeds left boundary point L 边界 When selecting L 边界 The location is used as the initial lateral reference position. When there is an obstacle at the right boundary point, the initial lateral reference position obtained from the above process can provide a reference for the vehicle path. By simply controlling the vehicle to be close to the initial lateral reference position, it can be ensured that the vehicle avoids the obstacle on the right, while also ensuring that the vehicle stays within the driving boundary of the lane.
[0125] Alternatively, in this embodiment, if there is an obstacle at the left boundary point, the corresponding left limit position is obtained based on the left boundary point corresponding to the obstacle. In this case, the initial lateral reference position can be expressed as: ref_l = clamp(L 极限 -buffer,R 边界,0), which indicates L 极限 - Given the condition that the buffer is located at the horizontal coordinates of the centerline and the right boundary point, L is preferred. 极限 -The horizontal coordinate of the buffer is the initial horizontal reference position; while when L 极限 - When the buffer position is to the left of the centerline, the position of the centerline is selected as the initial lateral reference position; when L 极限 - The buffer position is outside the right boundary point R. 边界 When selecting R 边界 The location is used as the initial lateral reference position. When there is an obstacle at the left boundary point, the initial lateral reference position obtained based on the above process can provide a reference for the vehicle path. By simply controlling the vehicle to be close to the initial lateral reference position, it can be ensured that the vehicle avoids the obstacle on the left, while also ensuring that the vehicle stays within the driving boundary of the lane.
[0126] C. When a non-offset driving decision is executed, and there are no obstacles at the left and right boundary points of the same longitudinal position of the lane, determine the distance between the right boundary point, the left boundary point and the vehicle respectively;
[0127] Obtain the pair of boundary points closest to the vehicle;
[0128] If the distance between the left and right boundary points at the same longitudinal position of the lane is less than a preset threshold, the initial lateral reference position is selected as the middle position between the current left and right boundary points. This stage indicates that both sides of the vehicle are close to the corresponding side boundary. At this time, in order to ensure that the vehicle does not exceed the driving boundary, the vehicle is controlled to be in the middle position of the driving boundary.
[0129] If the sum of the lateral coordinates of the right boundary point of the lane and the preset road margin is greater than 0, the initial lateral reference position is selected as the lateral position corresponding to the sum of the lateral coordinates of the right boundary point of the lane and the preset road margin. This stage indicates that the center line of the lane is close to the right boundary, and it is safer for the vehicle to drive to the left of the center line.
[0130] If the difference between the lateral coordinate of the left boundary point of the lane and the preset road margin is less than 0, the initial lateral reference position is selected as the lateral position corresponding to the difference between the lateral coordinate of the left boundary point of the lane and the preset road margin. This stage indicates that the center line of the lane is close to the left boundary of the lane, and it is safer for the vehicle to drive to the right of the center line.
[0131] In other cases, the initial lateral reference position is selected from the lateral position of the center line, i.e., driving close to the center.
[0132] Step Six:
[0133] Define the obstacle detection boundaries and obstacle containers for the vehicle, traverse obstacles in the longitudinal direction and classify them.
[0134] Specifically, the process of setting obstacle detection boundaries for the vehicle in this embodiment includes:
[0135] Obtain the vehicle's own body boundary and the road boundary of the lane; the vehicle body boundary includes the front boundary, rear boundary, left boundary and right boundary, and the road boundary includes the left road boundary and right road boundary;
[0136] The front obstacle detection boundary of the vehicle is calculated based on the front boundary of the vehicle and the preset forward detection distance.
[0137] The rear obstacle detection boundary of the vehicle is calculated based on the rear boundary of the vehicle and the preset rearward detection distance;
[0138] The first detection position is calculated based on the left boundary of the vehicle and the preset left detection distance, and the second detection position is calculated based on the left road boundary and the preset left road boundary margin; the position that is biased towards the center line between the first detection position and the second detection position is selected as the left obstacle detection boundary of the vehicle.
[0139] The third detection position is obtained by calculating the right boundary of the vehicle and the preset right detection distance, and the fourth detection position is obtained by calculating the right road boundary and the preset right road boundary margin. The position that is biased towards the center line between the third detection position and the fourth detection position is selected as the right obstacle detection boundary of the vehicle.
[0140] The obstacle detection boundary of the vehicle is obtained by merging the obtained front obstacle detection boundary, rear obstacle detection boundary, left obstacle detection boundary and right obstacle detection boundary.
[0141] In this embodiment, by defining the obstacle detection boundary of the vehicle and setting the vehicle's own detection range, obstacles within the range are included in the detection and subsequent analysis is performed, while obstacles outside the range are ignored, thereby reducing the computational load of vehicle obstacle detection and improving planning efficiency.
[0142] Furthermore, in this embodiment, the process of classifying obstacles is as follows:
[0143] First, create an empty left obstacle container and a right obstacle container respectively;
[0144] Subsequently, obstacles within the obstacle detection boundary of the vehicle are traversed longitudinally, and the boundaries of the obstacles are obtained. It should be noted that dynamic obstacles not within the obstacle detection boundary and dynamic obstacles moving in the opposite direction behind the vehicle in the longitudinal direction can be ignored in this stage. In addition, in this embodiment, multiple interfaces can be reserved for obstacles within the obstacle detection boundary so as to add obstacle types to be ignored as needed.
[0145] When the lateral coordinate of the left boundary of the obstacle is greater than the lateral coordinate of the vehicle's center axis, and the lateral coordinate of the right boundary of the obstacle is less than the lateral coordinate of the vehicle's center axis, it indicates that the obstacle occupies both sides of the vehicle obstacle detection boundary, such as stone blocks symmetrically arranged on both sides of the lane. At this time, the obstacle is simultaneously placed in the left obstacle container and the right obstacle container.
[0146] When the lateral coordinate of the left boundary of the obstacle is between the lateral coordinate of the right obstacle detection boundary and the lateral coordinate of the vehicle's center axis, it indicates that the obstacle occupies the right driving space of the vehicle, and the obstacle is placed into the right obstacle container.
[0147] When the lateral coordinate of the right boundary of the obstacle is located between the lateral coordinate of the left obstacle detection boundary and the lateral coordinate of the vehicle's center axis, the obstacle on its surface occupies the left driving space of the vehicle, and the obstacle is placed into the left obstacle container.
[0148] Step Seven:
[0149] Based on the obstacle classification results, obstacle avoidance decisions are matched for obstacles at different locations in the longitudinal direction, and lateral reference positions are calculated.
[0150] Furthermore, in this embodiment, the avoidance decision includes at least three types: KEEP indicates that the lateral reference position is taken from the lateral coordinate of the rear axle center of the vehicle, CENTER indicates that the lateral reference position is taken from the initial lateral reference position ref_l, and AVOID indicates that the lateral reference position is obtained by classifying and discussing the dynamic obstacle situation.
[0151] Specifically, when both the left and right obstacle containers of the vehicle are occupied, it indicates that the driving space on both sides of the vehicle is occupied and the vehicle should keep driving in the middle. At this time, the lateral reference position is selected from the lateral coordinates of the vehicle's center axis, i.e., the KEEP decision.
[0152] When neither the left nor right obstacle container of the vehicle is placed with an obstacle, it indicates that there is driving space on both sides of the vehicle. In this case, the lateral reference position is preferentially selected from the initial lateral reference position of the current lane, i.e., CENTER decision.
[0153] When only the left obstacle container contains obstacles, iterate through the left obstacle container and obtain the right boundary of the obstacle closest to the vehicle. Based on the right boundary of the obstacle, extend to the right by a first preset margin (usually twice the vehicle width) to obtain the first avoidance boundary obs_nudge_l; based on the right road boundary of the obstacle's current position, extend to the left by a second preset margin (usually 0.5 times the vehicle width) to obtain the second avoidance boundary right_limit_l.
[0154] The position that is biased towards the obstacle between the first avoidance boundary obs_nudge_l and the second avoidance boundary right_limit_l is selected as the first decision avoidance boundary, i.e., the AVOID decision.
[0155] The position furthest from the obstacle among the vehicle's centerline, the initial lateral reference position, and the first decision avoidance boundary is selected as the lateral reference position. In other words, the decision that is rightmost in the lateral coordinates of the three reference decisions (KEEP, CENTER, and AVOID) is selected as the final decision.
[0156] Alternatively, when only the right obstacle container contains obstacles, traverse the right obstacle container and obtain the left boundary of the obstacle closest to the vehicle. Based on the left boundary of the obstacle, extend the third preset margin to the left to obtain the third avoidance boundary obs_nudge_2; based on the left road boundary of the obstacle's current position, extend the fourth preset margin to the right to obtain the fourth avoidance boundary left_limit_l.
[0157] The position that is biased towards the obstacle between the third avoidance boundary obs_nudge_2 and the fourth avoidance boundary left_limit_l is selected as the second decision avoidance boundary, i.e., the AVOID decision.
[0158] The position furthest from the obstacle among the vehicle's centerline, the initial lateral reference position, and the second decision avoidance boundary is selected as the lateral reference position. That is, the leftmost decision in the lateral coordinates of the three reference decisions (KEEP, CENTER, and AVOID) is selected as the final decision.
[0159] Step 8:
[0160] The lateral reference positions at different locations are filtered to obtain the filtered lateral reference positions. The purpose of filtering the lateral reference positions in this step is to ensure the continuity of the changes in the lateral reference positions between frames; a first-order low-pass filter is sufficient here.
[0161] In addition, this application also discloses a vehicle comprising: a controller configured to perform the lateral reference position decision method described above.
[0162] Furthermore, this application also discloses a non-volatile storage medium, which includes a stored program, wherein the program, when running, controls the device where the non-volatile storage medium is located to execute the above-described lateral reference position decision method.
[0163] In addition, this application also discloses an electronic device comprising a processor and a memory; the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein the computer-readable instructions, when executed, perform the aforementioned lateral reference position decision method.
[0164] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable non-transitory storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0165] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0166] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0167] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0168] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.
Claims
1. A lateral reference position decision method for autonomous driving path planning, characterized in that, Includes the following steps: The driving boundaries and offset decisions within the lane are obtained, wherein the driving boundaries include at least the left boundary, the right boundary, and the centerline; and the offset decisions include at least offset driving decisions and non-offset driving decisions. A Frenet coordinate system is constructed based on the centerline of the lane, with the position of the centerline in the lateral direction as 0, the left side of the centerline as the positive direction, and the right side of the centerline as the negative direction; The left and right boundaries of the driving boundary are discretized longitudinally to obtain several longitudinally arranged boundary point pairs, each of which includes a left boundary point and a right boundary point. Traverse each pair of boundary points vertically to obtain the lateral limit positions corresponding to different pairs of boundary points. The lateral limit positions include the left limit position and the right limit position. The initial lateral reference position is obtained based on offset decision and lateral limit position analysis; Define the obstacle detection boundaries and obstacle containers for the vehicle, traverse obstacles in the longitudinal direction and classify them: Create empty left and right obstacle containers respectively; Traverse vertically the obstacles located within the obstacle detection boundary of the vehicle and obtain the boundaries of the obstacles; When the lateral coordinate of the left boundary of the obstacle is greater than the lateral coordinate of the vehicle's center axis, and the lateral coordinate of the right boundary of the obstacle is less than the lateral coordinate of the vehicle's center axis, the obstacle is simultaneously placed into the left obstacle container and the right obstacle container. When the lateral coordinate of the left boundary of the obstacle is between the lateral coordinate of the right obstacle detection boundary and the lateral coordinate of the vehicle's center axis, the obstacle is placed into the right obstacle container. When the lateral coordinate of the right boundary of the obstacle is between the lateral coordinate of the left obstacle detection boundary and the lateral coordinate of the vehicle's center axis, the obstacle is placed into the left obstacle container. Based on the obstacle classification results, obstacle avoidance decisions are made for obstacles at different locations in the longitudinal direction, and lateral reference positions are calculated. The lateral reference positions at different locations are filtered to obtain the filtered lateral reference positions.
2. The lateral reference position decision method according to claim 1, characterized in that, The steps for obtaining the initial lateral reference position based on offset decision and lateral limit position analysis include: When making an off-center driving decision, obtain the centerline of the lane and iterate through the curvature at each position on the centerline; The type of lane ahead of the vehicle is determined based on the curvature of the centerline, and the lane type includes at least a curve. When it is determined that there is a curve in front of the vehicle, the first reference position is calculated based on the right limit position and the preset limit margin; the position of the first reference position and the center line are determined, and the position that is more inclined to the right boundary point is selected as the initial lateral reference position.
3. The lateral reference position decision method according to claim 1, characterized in that, The steps for obtaining the initial lateral reference position based on offset decision and lateral limit position analysis include: When making an off-center driving decision, obtain the centerline of the lane and iterate through the curvature at each position on the centerline; The type of lane in front of the vehicle is determined based on the curvature of the centerline, and the lane type includes at least a straight road; When it is determined that the road in front of the vehicle is straight, check whether there are obstacles at the right boundary points of the lane; When there is an obstacle at any right boundary point, the first reference position is calculated based on the right limit position and the preset limit margin corresponding to the right boundary point. When the first reference position is located between the right limit position and the left boundary point in the same vertical position, the initial horizontal reference position is selected from the first reference position; When the first reference position is not located between the right limit position and the left boundary point in the same vertical position, the initial horizontal reference position is selected as the position that is closest to the first reference position between the right limit position and the left boundary point. When there are no obstacles at the right boundary points of the lane, the first reference position is calculated based on the right limit position and the preset limit margin. Determine the position of the first reference position relative to the center line. The initial lateral reference position is selected as the position closest to the right boundary point between the first reference position and the center line.
4. The lateral reference position decision method according to claim 1, characterized in that, The steps for obtaining the initial lateral reference position based on offset decision and lateral limit position analysis include: When a non-offset driving decision is executed, and there are obstacles at both the right and left boundary points of the same longitudinal position of the lane, the lateral distance between the obstacles corresponding to the right and left boundary points and the vehicle is determined respectively. Obtain the obstacle corresponding to the boundary point closest to the vehicle, and obtain the lateral limit position on the side where the obstacle is located; The first reference position is calculated based on the lateral limit position on the side where the obstacle is located and the preset limit margin. Analyze the first reference position, the lateral limit position corresponding to the boundary point on the opposite side of the obstacle, and the lateral coordinates of the centerline; When the first reference position is located between the lateral limit position on the opposite side of the obstacle and the centerline, the initial lateral reference position is selected from the first reference position; When the first reference position is not located between the lateral limit position and the center line on the opposite side of the obstacle, the initial lateral reference position is selected as the position closest to the first reference position between the lateral limit position and the center line on the opposite side of the obstacle.
5. The lateral reference position decision method according to claim 1, characterized in that, The steps for obtaining the initial lateral reference position based on offset decision and lateral limit position analysis include: When a non-offset driving decision is executed, and there is an obstacle at any boundary point of the right or left boundary point at the same longitudinal position of the lane, the first reference position is calculated based on the lateral limit position corresponding to the boundary point on the side where the obstacle is located and the preset limit margin. Analyze and determine the first reference position, the lateral limit position corresponding to the boundary point on the side where the obstacle is located, and the lateral coordinates of the centerline; When the first reference position is located between the lateral limit position on the side where the obstacle is located and the center line, the initial lateral reference position is selected as the first reference position; When the first reference position is not located between the lateral limit position and the center line on the side where the obstacle is located, the initial lateral reference position is selected as the position closest to the first reference position between the lateral limit position and the center line on the side where the obstacle is located.
6. The lateral reference position decision method according to claim 1, characterized in that, The steps for obtaining the initial lateral reference position based on offset decision and lateral limit position analysis include: When a non-offset driving decision is executed, and there are no obstacles at the left and right boundary points of the same longitudinal position of the lane, the distances between the right and left boundary points and the vehicle are determined respectively. If the distance between the left and right boundary points at the same longitudinal position of the lane is less than a preset threshold, the initial lateral reference position is selected as the middle position between the current left and right boundary points; If the sum of the lateral coordinates of the right boundary point of the lane and the preset road margin is greater than 0, the initial lateral reference position is selected as the lateral position corresponding to the sum of the lateral coordinates of the right boundary point of the lane and the preset road margin. If the difference between the lateral coordinate of the left boundary point of the lane and the preset road margin is less than 0, the initial lateral reference position is selected as the lateral position corresponding to the difference between the lateral coordinate of the left boundary point of the lane and the preset road margin. Otherwise, the initial lateral reference position is selected from the lateral position where the center line is located.
7. The lateral reference position decision method according to claim 1, characterized in that, The steps for setting obstacle detection boundaries for the vehicle include: Obtain the vehicle's own body boundary and the road boundary of the lane; the vehicle body boundary includes the front boundary, rear boundary, left boundary and right boundary, and the road boundary includes the left road boundary and right road boundary; The front obstacle detection boundary of the vehicle is calculated based on the front boundary of the vehicle and the preset forward detection distance. The rear obstacle detection boundary of the vehicle is calculated based on the rear boundary of the vehicle and the preset rearward detection distance; The first detection position is calculated based on the left boundary of the vehicle and the preset left detection distance, and the second detection position is calculated based on the left road boundary and the preset left road boundary margin; the position that is biased towards the center line between the first detection position and the second detection position is selected as the left obstacle detection boundary of the vehicle. The third detection position is obtained by calculating the right boundary of the vehicle and the preset right detection distance, and the fourth detection position is obtained by calculating the right road boundary and the preset right road boundary margin. The position that is biased towards the center line between the third detection position and the fourth detection position is selected as the right obstacle detection boundary of the vehicle. The obstacle detection boundary of the vehicle is obtained by merging the obtained front obstacle detection boundary, rear obstacle detection boundary, left obstacle detection boundary and right obstacle detection boundary.
8. The lateral reference position decision method according to claim 7, characterized in that, The steps of matching and avoiding obstacles at different positions in the longitudinal direction based on the obstacle classification results and calculating the lateral reference position include: When both the left and right obstacle containers of the vehicle are filled with obstacles, the lateral reference position is selected from the lateral coordinates of the vehicle's center axis. When neither the left nor right obstacle container of the vehicle is occupied by an obstacle, the lateral reference position is selected as the initial lateral reference position of the current lane. When only the left obstacle container contains an obstacle, traverse the left obstacle container and obtain the right boundary of the obstacle closest to the vehicle. Based on the right boundary of the obstacle, extend to the right by a first preset margin to obtain the first avoidance boundary; based on the right road boundary of the obstacle's current position, extend to the left by a second preset margin to obtain the second avoidance boundary. The position that is biased towards the obstacle between the first avoidance boundary and the second avoidance boundary is selected as the first decision avoidance boundary. The position furthest from the obstacle among the vehicle's centerline, the initial lateral reference position, and the first decision avoidance boundary is selected as the lateral reference position. When only the right obstacle container contains an obstacle, traverse the right obstacle container and obtain the left boundary of the obstacle closest to the vehicle. Based on the left boundary of the obstacle, extend to the left by a third preset margin to obtain the third avoidance boundary; based on the left road boundary of the obstacle's current position, extend to the right by a fourth preset margin to obtain the fourth avoidance boundary. The position that is biased towards the obstacle between the third and fourth avoidance boundaries is selected as the second decision avoidance boundary; The position furthest from the obstacle among the vehicle's centerline, the initial lateral reference position, and the second decision avoidance boundary is selected as the lateral reference position.
9. A vehicle, characterized in that, The vehicle includes a controller configured to perform the lateral reference position decision method according to any one of claims 1-8.